Artificial intelligence in artistic creation:

Artificial intelligence in artistic creation: Much ado about nothing or a real paradigm shift?

By Yan St-Onge, AI Project (Chantier IA) documentation officer

** This text was the subject of a presentation in Montreal as part of the Symposium “AI Companionability / Compagnonnage IA” on March 6, 2025 at Hexagram.

INTRODUCTION

I work with artists in residence at Sporobole as part of the Chantier IA, trying to understand the impact and changes in creative practices. So today I’m going to share with you a few elements of my research since last year on the challenges of using AI in an artistic process.

The rhythm and time of creation

Some steps are quicker when working with AI, but others take much longer, as it often takes a lot of trial and error before something satisfying is produced.

Several projects that were done in residence in 2024 at Sporobole relied on the use of many different AIs. For example, two or three AIs are used to generate images, another AI to do upscaling (a quality enhancement), another to generate video from the images, another for sound, and through it all, one or two LLM-type AIs to discuss the project and seek solutions with a conversational agent.

So far, the industry marketing that makes us believe that generative AI enables instant creation is anything but realistic when you look at how artists are integrating AI into their creative process.

For video creation, it’s only possible to generate a few seconds at a time, which makes the creation of a medium or feature-length film very tedious, at least for the time being.

The quantity of images/text/sound produced

The issue of image quantity has existed since the advent of digital photography, which is much faster and cheaper than film photography. It also applies to video and music, where the switch to digital means that content production is no longer limited by material constraints. With AI, the issue of quantity becomes even more striking.

One wonders whether the artist isn’t becoming an artist-curator of his own work. Indeed, it’s not so easy to choose what’s worthy of interest when you’re producing content at high speed. How do you choose one image from thousands of similar images with slight variations?

The act of creating is shifting to a task of selection.

Rapid evolution of AI tools

In 2024, artists found it difficult to keep up and adapt to the new AI tools that were being released at a very rapid pace. The danger is spending more time trying to learn how to use new tools than spending time creating. For example, the options for using AI in music were far fewer and less interesting than those in image and video generation, a reality that makes artists even more eager to try out a new application or fresh AI model.

Representation bias

AIs are trained on pre-existing content. As a result, the biases and stereotypes that exist in society are unfortunately amplified by generative AI. It’s common to come across ethnic, racist, sexist or homophobic biases in AI-generated content.

The image style options offered by most AIs often have very definite aesthetics that are hardly suitable for an artistic approach where innovation and creativity play a big role.

Although this problem tends to diminish over time, AI has difficulty accurately representing the human body, particularly hands, fingers and legs.

Finally, a bias that undeniably affects freedom of expression is the moral and political censorship of artificial intelligence tools. While the Chinese DeepSeek application censors parts of China’s political history, such as the Tiananmen Square protests, Western tools themselves have blocks on certain forms of representation, whether erotic, violent or hateful in nature.

Control and loss of control

At what point do we, as artists, agree to let algorithms decide? When we agree to give a share of intentionality to the machine for a project, is the possibility of innovation great?

Although many artists accept or even wish to delegate a share of artistic decisions to artificial intelligence, the balance between control and loss of control is often not easy to find.

Artistic value

The media treatment of AI generally gives a negative image of its use, which is associated with cheating in schools or job losses in many fields. As a result, artists face a dilemma: to clearly state their use of AI, which can lead to very trenchant positions from the public and people in the arts community, or, not to state it, but risk being reproached for it.

We probably need to distinguish between the gesture of GENERATING an image with AI and the fact of CREATING an image with AI. The first verb evokes a gesture, a step, while the second evokes a process, in which there are several successive stages of generation. Automated generation by AI is not an end in itself, but a means that can be used to create. So, contrary to the myth propagated by the media, AI does not enable us to create a work of art automatically; it is a tool that must support an artistic approach.

Towards a redefinition of art

My title was intended to be a little caricatural. In fact, AI may well have the effect of a paradigm shift in the history of art, it remains to be seen how far-reaching. The impact is as much on the creative work itself as on aesthetic choices, but also on various aspects of cultural work. The change that is taking place is also ontological, since the popularization of AI is forcing us to question what art fundamentally is. We need to collectively reflect on the distinction between an image and a work of art. Just as photography, collage, the readymade and, more recently, conceptual art have had an undeniable impact on our conception of what a work of art is, AI is forcing us to reflect on and redefine art. The entire art world, including the public, must adapt to this reality, as well as to the changes that artificial intelligence will have on other media, even the most traditional.

Like any new technology, AI will have its advantages and disadvantages. There is and will continue to be a movement of rejection and resistance, which will lead to a return of value to more traditional and artisanal artistic practices. But who knows if there won’t also be new art forms or a new artistic movement, in which generative AI will occupy a prominent place.